Skip to main content

Welcome to a Grasp lesson

Create a free account to keep studying and create your own study path.

Lesson 6

Limits of Conclusions in Aphantasia Group-Comparison Studies

Recommended resources

Hello. The previous lesson used actual-light and mental-effort controls to interpret a pupil-response experiment. Those controls strengthened a specific comparison, but they did not turn an average difference between groups into a claim about every person with aphantasia. This lesson examines that gap. You will use a study of reported imagery, memory, and dreaming to judge which conclusions its recruitment, sample size, and group comparisons support—and which they do not.


Start with who entered the study

A group-comparison study asks whether a measured outcome differs, on average, between groups assembled in a particular way. Before interpreting a difference, find out who was eligible, who volunteered, and who served as the comparison.

Read the Participants part of “A cognitive profile of multi-sensory imagery, memory and dreaming in aphantasia.” It gives enough detail to evaluate the groups without needing to work through the paper’s statistical tests.

A cognitive profile of multi-sensory imagery, memory and dreaming in aphantasia

Read the paper’s Methods section to see how its groups were assembled and how the authors planned their sample size.

Under “Method” and then “Participants,” read the participant account through the end of “Control group 2 (Undergraduates).” Pay particular attention to “Aphantasia group,” “Control group 1 (MTurk),” and “Control group 2 (Undergraduates)”: compare how each group was recruited, not just how many people it contained.

The aphantasia participants identified themselves as aphantasic and came from an online research community. One comparison group came from a paid online participant pool; the other comprised psychology students receiving course credit. These are three different routes into the study. Someone who has learned about aphantasia and seeks out related research may differ from a paid survey participant in more ways than imagery experience—for example, in interest in mental experience or familiarity with the term.

That is a selection concern, not a reason to discard the findings. The study can describe differences among the people who took part. But its aphantasia sample is not a random sample of everyone with aphantasia. Consequently, a group average here cannot establish how all people with aphantasia remember or dream, nor can the proportion of self-identified participants tell us how common aphantasia is.

For a short research perspective on this problem, watch the following segment. Treat it as an explanation of a recruitment challenge, rather than as evidence for a particular memory or dreaming result.

Scientists Discover Why Some People Can't Visualize | Aphantasia Research 2025

In “Scientists Discover Why Some People Can't Visualize | Aphantasia Research 2025,” Aphantasia Network’s interview with Dr. Spagna identifies a limitation shared by much aphantasia research.

Watch recruitment limits. Focus on why researchers readily encounter people who already know they have unusual imagery, while people with similar experiences who have never encountered the term are harder to recruit.



A larger sample solves some problems, not every problem

The study aimed for at least 170 people in each comparison group and obtained 267 self-identified aphantasic participants, alongside comparison groups of roughly 200 each. That is substantially more informative about average differences within these recruited groups than a comparison involving just a handful of volunteers. Its planning also accounted for the possibility that earlier, smaller studies had made group differences look larger than they were.

Yet sample size and representativeness answer different questions. More participants can help estimate a difference more precisely and detect smaller differences. It cannot, by itself, remove a bias in who tends to volunteer. If recruitment repeatedly attracts people with the same unmeasured characteristic, recruiting more of them may produce a very precise estimate for a group that still does not represent the wider population.

A prevalence study makes this distinction especially clear. Read how another research team planned a survey of imagery ability and chose its recruitment wording.

An international estimate of the prevalence of differing visual imagery abilities

Read this study as a contrast in research design: estimating how common an imagery category is requires different sampling decisions from comparing outcomes in groups already recruited for their imagery status.

In Methods, find the sample-size paragraph immediately before “Materials.” Read the calculation, noting that its target concerns precision for a prevalence estimate. Then find the Methods paragraph beginning “Members of the research team distributed the link or QR code” and read the recruitment description to the end of that paragraph. Notice what the invitation deliberately did not mention.

That team calculated a target sample size of 1,969 for its desired precision in estimating prevalence and avoided mentioning imagery or aphantasia in recruitment announcements. Omitting those terms was intended to reduce the chance that people aware of unusual imagery would be especially likely—or unlikely—to participate. It does not make recruitment perfectly random or eliminate all response bias. It does show why a sample drawn from an aphantasia research community and a survey designed to estimate prevalence cannot answer the same population-level question.



A second control group is a check, not a cure

The first comparison group in the cognitive-profile study was matched to the aphantasia group on mean age, but the groups differed in their proportions of women and men. The second comparison group provided another check, with a different gender composition—but its student participants were, on average, considerably younger than the aphantasia group. Matching one characteristic did not match them on everything else.

Read the authors’ comparison of the two control groups and their account of the study’s limitations. The numbers attached to individual questionnaire items are less important here than whether a finding persists across comparisons.

A cognitive profile of multi-sensory imagery, memory and dreaming in aphantasia

Return to the paper to see what the second comparison checks, what changes between comparisons, and what remains uncertain.

In Results, under “Control Group 2: Replication Analysis,” read the comparison rationale. Then, in “Discussion,” read the demographic limits. Finally, stay in “Discussion” and read the measurement limits. Focus on why a broadly similar pattern across two controls is useful without making either control perfectly matched.

Many imagery- and memory-related self-report differences appeared in both comparisons. That persistence makes a simple explanation tied only to the first control group less convincing. But some results—particularly concerning mind-wandering and aspects of dreams—differed depending on which control group was used. The younger age of the student group offers one possible explanation. The second comparison reused the same aphantasia group; it was not a fresh replication with independently recruited participants in both groups.

Measurement sets another boundary. The paper asked people about their experiences and perceived abilities. A lower rating of remembered detail is evidence about reported recollection, not, on its own, proof that someone failed to retain an event or performed worse on an objective memory task. The authors note that response tendencies cannot be ruled out entirely, while also pointing out that participants did not simply report low ability on every scale. Both observations belong in a fair reading.

Most importantly, participants were not randomly assigned to have or lack voluntary visual imagery. Even if a group difference is statistically convincing, this design establishes an association in these samples. It cannot establish that imagery absence caused a difference in memory reports or dream reports. Differences in recruitment, other characteristics of participants, and how experiences are described could also contribute. Two imperfectly matched control groups help test some alternatives; they cannot exclude every alternative.

A careful conclusion therefore names the group, measure, and comparison: In this recruited sample, self-identified aphantasic participants reported lower scores on certain recollection measures than the comparison participants. “Aphantasia makes people unable to remember” changes the measure from a report to a general ability, changes an average to a universal claim, and changes an association to a cause. None of those changes is justified by this design.



What to carry forward

When you encounter a group-comparison claim, check three boundaries:

  • Recruitment: Do the volunteers represent the population the claim is about?
  • Sample size: Is there enough data for the intended comparison—and could a larger sample still retain selection bias?
  • Inference: Does the result describe a measured association, or is someone presenting it as a cause, a universal trait, or an objectively measured ability?

These questions do not make the study’s differences uninteresting. They make its contribution more precise. The next module applies that precision to memory, beginning with the distinction between recalling a personal episode, knowing a fact, and recognizing something encountered before.